Just days before the U.S. presidential election, “positive sentiment” for Donald Trump was soaring nationwide across all the major social media platforms. The Donald was crushing it online while official polls showed an assured Clinton victory.
Trump’s “surprise” victory in the U.S. presidential election validates the point we’ve been making at Buzz Indexes since we opened shop: traditional polling and survey methods are out-of-date and, essentially, useless.
Social media sentiment is the way to go if you want meaningful opinion data. Here’s why:
- Traditional polls and surveys are full of bugs. As sources of opinion data they suffer from human behavioral weaknesses that handicap their effectiveness. These include:
- Response bias — how people respond to survey questions is affected by more than just their beliefs about the answers. The way questions are phrased; the looks or manner of the person conducting the survey; the polling method; and even the participant’s desire to have “good” responses; all these can affect the survey results.
- Forced choices — a style of survey format that limits response options so they fall neatly into two (sometimes three) categories. But limiting the range of acceptable choices can bias results because it forces respondents to choose between two unacceptable alternatives or the third, neutral “undecided” response. With forced choice questions, respondents are unable to show they dislike the response options. Instead, they’re forced into a choice that doesn’t fully capture their view.
- Yea-saying — this is the tendency of survey respondents to agree with whatever statement is presented to them. It’s an effect often caused by interviewers being overly friendly when asking survey questions.
- Low incentive for truth telling — most surveys and polls are anonymous, which means respondents can give a false answer without risk to their reputations. There is no immediate penalty or terrible consequence for answering dishonestly to a survey question, so there’s no incentive to tell the truth. People may even have an incentive to lie by taking pleasure in purposefully skewing results.
- Biased sampling is a thing. Pollsters are human, which means they are prone to errors and (let’s be honest), laziness. In fact, “convenience sampling” is the most common bias. It occurs when pollsters use a convenient strategy (such as polling people within a certain zip code) instead of a truly random selection process from a massive database of names. Biased sampling skews the data toward a certain demographic so it’s impossible to extrapolate results to the larger population.
- Insights from a crowd contain (often hidden) truths. Trump’s presidential win illustrates the misleading nature of traditional polls and surveys. In fact, social and polling data showed opposing results. When Trump was ahead on Facebook and Twitter, he was behind in the polls. Scandals that cost him in the polls actually increased his social media engagement. And if you analyze social media chatter about the election, it’s clear that the Trump campaign had more traction than the Clinton campaign over the long haul. One example comes from the social media analytics firm Crowdbabble, which tracked engagement from Clinton and Trump supporters on Twitter, Facebook, and Instagram from January 1 to November 6 this year. “Engagement” means interacting with or endorsing content posted by the two candidates. The results show Trump garnered 351 million engagements in total, versus Clinton’s 145 million.
- Social media and the sentiment factor. Unlike polls and surveys, which are prone to human bias and error, social media provides the ideal platform for good insights from the crowd. Here’s why:
- Diversity of opinion—the hundreds of millions of people who use social media platforms worldwide are highly likely to have their own individual ideas based on personal thinking.
- Independence of ideas—the individual opinions of people on social media are less likely to be influenced by the opinions of others or succumb to groupthink.
- Diversity of participants—social media activity is a global phenomenon and participants are widely distributed, which means they have access to distinctive local knowledge that informs their opinions.
- Number of participants—Facebook, Twitter, Instagram and other platforms are all massive sample sizes for data gatherers. If you’re an investor, imagine the benefits of harnessing the power of social media’s Big Data to measure sentiment about stock prices.
- High incentive for truth telling—fear of being called out (or being unfollowed) keeps most people on social media truthful in what they post. People naturally seek validation from peers. We know that authenticity wins more kudos in the form of shares, likes, retweets, and comments.
Stay on the smart track when it comes to your investing, with crowd-sourced sentiment made available by the BUZZ Indexes ETF.